主挂了之后导致服务2分钟时间不可用

线上环境基于基于codis3.2的版本，单个分区的主挂了以后，需要2分钟时间才能完全恢复。
下面是sentinel的日志
`
7439:X 24 Apr 13:56:50.455 # +odown master vir-codis-1 10.100.6.89 6379 #quorum 2/2
7439:X 24 Apr 13:56:50.455 # +new-epoch 10
7439:X 24 Apr 13:56:50.455 # +try-failover master vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:50.460 # +vote-for-leader e15eb691d9140a1de300ff3d5822ff3121d40239 10
7439:X 24 Apr 13:56:50.465 # dd8a158fbef73ab5332e041edfffd7674cf465cb voted for e15eb691d9140a1de300ff3d5822ff3121d40239 10
7439:X 24 Apr 13:56:50.465 # 2c181af837c75b004445d826658687ef8d2e5ae7 voted for e15eb691d9140a1de300ff3d5822ff3121d40239 10
7439:X 24 Apr 13:56:50.561 # +elected-leader master vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:50.561 # +failover-state-select-slave master vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:50.619 # +selected-slave slave 10.100.6.95:6379 10.100.6.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:50.619 * +failover-state-send-slaveof-noone slave 10.100.6.95:6379 10.100.6.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:50.690 * +failover-state-wait-promotion slave 10.100.6.95:6379 10.100.6.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:51.508 # +promoted-slave slave 10.100.6.95:6379 10.100.6.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:51.508 # +failover-state-reconf-slaves master vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:51.587 * +slave-reconf-sent slave 10.100.2.95:6379 10.100.2.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:52.547 * +slave-reconf-inprog slave 10.100.2.95:6379 10.100.2.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:56:52.648 # -odown master vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:58:48.710 * +slave-reconf-done slave 10.100.2.95:6379 10.100.2.95 6379 @ vir-codis-1 10.100.6.89 6379
7439:X 24 Apr 13:58:48.810 # +failover-end master vir-codis-1 10.100.6.89 6379
`
发现时间花在了“slave-reconf-done”上面，请问有什么办法减少故障恢复时间吗？